[Trac #1511] Roll off low frequency drive to speakers

[Zarro Boogs per Child]

#1511: Roll off low frequency drive to speakers
-------------------------------+--------------------------------------------
 Reporter:  wmb at firmworks.com  |       Owner:  mlj      
     Type:  defect             |      Status:  new      
 Priority:  normal             |   Milestone:  Untriaged
Component:  hardware           |     Version:           
 Keywords:                     |    Verified:  0        
-------------------------------+--------------------------------------------
 I have been measuring the characteristics of the internal speakers.  They
 start to roll off at the low end somewhere in the vicinity of 600 Hz.  By
 350 Hz they are completely inaudible.

 This isn't surprising, considering that they are so small - it takes a big
 cone to move air at low frequencies.  Furthermore, the speakers are very
 shallow front-to-back, so the cones cannot move very far.

 The capacitors that couple the signal from the CODEC to the amplifier,
 C362, C363, C402, and C403, are 0.1uF.  With the amplifier's differential
 Zin of 150 K, that puts the corner frequency at about 20 Hz.

 So, for audio signals with low frequency energy, the speakers are seeing
 drive signals in the range from 20 Hz to about 600 Hz from which it can't
 produce useful sound.  However, the voice coils can conduct such
 frequencies and produce magnetic fields to drive the cones.  That
 negatively affects the ability of the speakers to reproduce the
 frequencies that they can handle.  When a low frequency signal drives a
 voice coil to near the end of its range, it can go no further, so high
 frequency signals that are present at the same time just get distorted.

 We could reduce that effect and make the system sound better by changing
 the capacitor values from 0.1 uF to 0.003 uF.  This would have no effect
 on the headphone frequency response.  It only affects the built-in
 speakers.

 Ideally, it would be better to have an even sharper cutoff filter, but
 that would require more components.  The single-pole rolloff of the
 coupling caps against the input impedance will help, and will cost
 nothing.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1511>
One Laptop Per Child <http://laptop.org/>